- Email: [email protected]

Contents lists available at ScienceDirect

European Journal of Combinatorics journal homepage: www.elsevier.com/locate/ejc

Imprimitive symmetric graphs with cyclic blocks Cai Heng Li a , Cheryl E. Praeger a , Sanming Zhou b a

School of Mathematics and Statistics, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia

b

Department of Mathematics and Statistics, The University of Melbourne, Parkville, VIC 3010, Australia

article

info

Article history: Received 27 September 2007 Accepted 17 February 2009 Available online 7 April 2009

abstract Let Γ be a graph admitting an arc-transitive subgroup G of automorphisms that leaves invariant a vertex partition B with parts of size v ≥ 3. In this paper we study such graphs where: for B, C ∈ B connected by some edge of Γ , exactly two vertices of B lie on no edge with a vertex of C ; and as C runs over all parts of B connected to B these vertex pairs (ignoring multiplicities) form a cycle. We prove that this occurs if and only if v = 3 or 4, and moreover we give three geometric or group theoretic constructions of infinite families of such graphs. © 2009 Elsevier Ltd. All rights reserved.

1. Introduction A graph Γ = (V , E ) is G-symmetric if G ≤ Aut(Γ ) is transitive on the set Arc(Γ ) of arcs of Γ , where an arc is an ordered pair of adjacent vertices. For a G-symmetric graph Γ , a partition B of V is G-invariant if B ∈ B implies Bg ∈ B for all g ∈ G, where Bg = {α g : α ∈ B}, and B is nontrivial if 1 < |B| < |V |. Such a vertex partition gives rise to a quotient graph ΓB , namely the graph with vertex set B in which B, C ∈ B are adjacent if and only if there exists an edge of Γ joining a vertex of B to a vertex of C . Since Γ is G-symmetric and B is G-invariant, ΓB is G-symmetric under the induced (not necessarily faithful) action of G on B . Moreover, if Γ is connected, then ΓB is connected and in particular all arcs join distinct parts of B . For an arc (B, C ) of ΓB , the subgraph Γ [B, C ] of Γ induced on B ∪ C with isolated vertices deleted is bipartite and, up to isomorphism, is independent of (B, C ). In some examples, such as the case where Γ is a cover of ΓB , all vertices of B and C occur in Γ [B, C ], but many other possibilities also arise. S For an arc (B, C ) of ΓB , let Γ (C ) = α∈C Γ (α), where Γ (α) denotes the set of vertices adjacent to α in Γ , and set

v := |B|,

k := |Γ (C ) ∩ B|.

(1)

E-mail addresses: [email protected] (C.H. Li), [email protected] (C.E. Praeger), [email protected] (S. Zhou). 0195-6698/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejc.2009.02.006

C.H. Li et al. / European Journal of Combinatorics 31 (2010) 362–367

363

An approach to understanding general G-symmetric graphs Γ in terms of ΓB , Γ [B, C ] and a 1-design induced on B was suggested in [3], and developed further in [5,8,9] in the case k = v − 1, where special additional structure on the parts B can be defined and exploited. If k = v − 2 it turns out that we may also define additional structure on the parts. Since Γ [B, C ] consists of k vertices from each of B and C , in particular v = k + 2 ≥ 3, and the set B \ Γ (C ) contains exactly two vertices. Thus we may define a multigraph Γ B with vertex set B and an edge joining the two vertices of B \ Γ (C ) for each C in the set ΓB (B) of parts of B adjacent to B in ΓB . Denote by Simple(Γ B ) the underlying simple graph of Γ B . It was proved [4, Theorem 2.1] that Simple(Γ B ) is GB -vertex-transitive and GB -edge-transitive, and either Γ B is connected or Simple(Γ B ) is a perfect matching (v/2) · K2 , where GB is the setwise stabiliser of B in G. In the latter case detailed information about Γ was obtained in [4, Theorem 1.3] when Γ B is simple. However, no information about Γ was obtained in the case where Γ B is connected. Here we considered the simplest possibility, namely Simple(Γ B ) has valency two. We find with surprise that the parts of B must have size 3 or 4 in this case. Our main result is Theorem 1.1 below. It involves the multiplicity m of the edges of the multigraph Γ B , that is, for a pair {α, β} of adjacent vertices of Γ B , m = |{C ∈ ΓB (B) : B \ Γ (C ) = {α, β}}|. Theorem 1.1. Suppose Γ is a G-symmetric graph (where G ≤ Aut(Γ )) whose vertex set admits a nontrivial G-invariant partition B such that k = v − 2 ≥ 1 with k, v as in (1), ΓB is connected, and Simple(Γ B ) has valency two. Then Simple(Γ B ) = Cv , ΓB has valency mv , and one of the following (a)–(c) occurs for an arc (B, C ) of ΓB . (a) v = 3 and Γ has valency m; (b) v = 4, Γ [B, C ] = K2,2 , and Γ is connected of valency 4m; (c) v = 4, Γ [B, C ] = 2 · K2 , and Γ has valency 2m. Remark 1.2. (1) In particular, if Γ B is simple, then in case (a) we have Γ = (|V (Γ )|/2) · K2 , and, in case (c), Γ has valency two and hence is a vertex-disjoint union of cycles of the same length. In Section 3 we construct an infinite family of graphs for each of these cases, and an infinite family of graphs for case (b) with Γ B simple by using the coset graph construction. (2) In cases (b) and (c) we prove that GBB ∼ = D8 , and for an arc (B, C ) of ΓB , we prove that B∪C ∼ GBC = Z2 × Z2 in case (b), and Z2 in case (c). (3) In case (a), Γ can be (G, 2)-arc-transitive even when m > 1; see [4, Example 4.6] for an infinite family of such graphs. In case (b) it is clear that Γ is not (G, 2)-arc-transitive. In case (c), if m > 1, then the stabiliser Gα of α in G is imprimitive on Γ (α) and hence Γ is not (G, 2)-arc-transitive. An example for case (c) such that Γ is (G, 2)-arc-transitive (hence m = 1) can be found in [4, Example 4.7]. Our construction for case (b) leads to an infinite family of connected 4-valent symmetric graphs Γ which have a 4-valent quotient not covered by Γ . To the best of our knowledge this is the first infinite family of symmetric graphs with these properties. Corollary 1.3. There exists an infinite family of connected symmetric graphs Γ of valency 4 which have a quotient graph ΓB of valency 4 such that Γ is not a cover of ΓB . In the light of Theorem 1.1 we ask, for other connected graphs Simple(Γ B ): Question 1.4. In the case where k = v − 2 and Γ B is connected, is v bounded by some function of the valency of Simple(Γ B )? We may also ask the following question. Question 1.5. Can Γ in Theorem 1.1 be determined for small values of m? The proof of Theorem 1.1 is given in Section 2 and the examples are constructed in Section 3. The reader is referred to [1] for group theoretic terminology used in the paper.

364

C.H. Li et al. / European Journal of Combinatorics 31 (2010) 362–367

2. Proof of Theorem 1.1 Two parts B, C ∈ B are called adjacent if they are adjacent in the quotient graph ΓB , and if B, C are adjacent we write GBC = (GB )C and let e(B, C ) be the edge of Simple(Γ B ) joining the two vertices of B \ Γ (C ). Proof of Theorem 1.1. Let (Γ , G, B ) satisfy the conditions of Theorem 1.1, and let B, C be adjacent parts. Then Simple(Γ B ) 6= (v/2)· K2 since it is of valency two by our assumption. Thus Γ B , and hence also Simple(Γ B ), is connected [4, Theorem 2.1] and so Simple(Γ B ) = Cv . Thus, by the definition of Γ B , the valency of ΓB is mv . Case 1: v odd. Since v is odd, there exists a unique vertex α ∈ B which is ‘antipodal’ to the edge e(B, C ) of Simple(Γ B ), that is, α is the unique vertex equi-distant in Simple(Γ B ) from the two vertices of e(B, C ). Now each element of GBC fixes B \ Γ (C ) setwise and hence fixes α . (In the case where m > 1, an element of GBC may permute the m edges of Γ B joining the two vertices of B \ Γ (C ).) Thus, GBC ≤ Gα . Since α 6∈ B \ Γ (C ), there exists β ∈ C adjacent to α in Γ . Suppose v ≥ 5. Then there exists a vertex γ ∈ B such that γ 6∈ {α} ∪ (B \ Γ (C )) and so γ is adjacent to a vertex δ ∈ C . Since Γ is G-symmetric, there exists g ∈ G such that (α, β)g = (γ , δ). Since g maps α ∈ B to γ ∈ B and β ∈ C to δ ∈ C , it fixes B and C setwise. Thus, g ∈ GBC ≤ Gα , which is a contradiction since α g = γ 6= α . Therefore, v = 3 and consequently Γ has valency m. Case 2: v even. Since v is even, there exists a unique edge of Simple(Γ B ), say, e = {α, β}, which is ‘antipodal’ to e(B, C ) in Simple(Γ B ), that is, α and β are both at maximum distance v/2 from some vertex of e(B, C ). Note that α, β ∈ B ∩ Γ (C ). Each vertex γ ∈ B ∩ Γ (C ) is adjacent to some vertex δγ ∈ C . Since Γ is G-symmetric, for each such γ there exists gγ ∈ G such that (α, δα )gγ = (γ , δγ ). Since gγ maps α ∈ B to γ ∈ B and δα ∈ C to δγ ∈ C , we have gγ ∈ GBC . Thus for each γ ∈ B ∩ Γ (C ), gγ fixes e(B, C ) setwise and hence fixes e = {α, β} setwise also. Thus α gγ = γ ∈ {α, β} and in particular v = 4 and B ∩ Γ (C ) = {α, β}. Since gβ fixes e setwise, it interchanges α and β . Since GBB is transitive on B, it follows that GBB ∼ = D8 . Therefore, 1 6= GBBC∪C ≤ hxB i × hxC i, where xB is the B C reflection of Γ in e(B, C ) and x is the reflection of Γ C in e(C , B). Note that xB interchanges α and β since it interchanges the two vertices of e(B, C ). Thus gβB = xB . Similarly, xC interchanges the two vertices of C \ e(C , B) = {η, ζ }, say, and GBC contains an element h such that hC = xC . Thus either GBBC∪C = hxB i × hxC i or GBBC∪C = hxB xC i ∼ = Z2 . Since GBC preserves the adjacency of Γ , the first possibility occurs if and only if α is adjacent to both of the vertices of C \ e(C , B) and hence Γ [B, C ] = K2,2 is the 4cycle (α, η, β, ζ , α). Since Simple(Γ B ) = C4 , in this case α is at distance 2 in Γ from β and from one of the vertices of e(B, C ), and at distance 3 or 4 from the other vertex of e(B, C ). Since ΓB is connected, it B∪C follows that Γ is connected of valency 4m in this case. Suppose now that GBC = hxB xC i ∼ = Z2 . Then the bipartite graph Γ [B, C ] consists of two edges only, namely, {α, δα } and {β, δβ }. Hence Γ [B, C ] = 2 · K2 and Γ has valency 2m. 3. Constructions In this section we present several constructions of infinite families of graphs that satisfy the conditions of Theorem 1.1 in the case where the multigraph Γ B is simple, that is Γ B = Simple(Γ B ) or equivalently m = 1. The first two constructions involve regular maps on surfaces. Here and in what follows our use of the term ‘regular map’ agrees with that of [2], that is, a regular map is a 2cell embedding of a connected (multi)graph on a closed surface such that its automorphism group is regular on incident vertex–edge–face triples. 3.1. Truncations of trivalent symmetric graphs The construction below produces all graphs that arise in case (a) of Theorem 1.1 with m = 1. Construction 3.1. Let Σ be a trivalent G-symmetric graph with n edges. Define Γ (Σ ) to be the graph with vertex set Arc(Σ ) and edges {(σ , τ ), (τ , σ )} for (σ , τ ) ∈ Arc(Σ ) [4, Example 2.4]. Then Γ (Σ ) = n · K2 , Γ (Σ ) is G-symmetric, and its vertex set admits the G-invariant partition

C.H. Li et al. / European Journal of Combinatorics 31 (2010) 362–367

365

Fig. 1. Obtaining Γ = 6 · K2 (heavy edges in (b)) by truncating the tetrahedron as in (a).

B (Σ ) = {B(σ ) : σ ∈ V (Σ )} with parts of size v = 3, where B(σ ) is the set of arcs of Σ with first vertex σ . For this partition we have k = v − 2 = 1, Γ (Σ )B(σ ) is the simple graph C3 , and Σ is isomorphic to the quotient graph Γ (Σ )B (Σ ) via the bijection σ 7→ B(σ ). As explained in [4, Example 2.4] this construction produces all imprimitive G-symmetric graphs

(Γ , B ) such that k = v − 2 = 1 and Γ B = C3 is simple. In case (a) of Theorem 1.1, if m = 1, then GBB ∼ = Z3 or D6 . From [2, Theorem 1.1] the former occurs if and only if ΓB admits an embedding as an orientably-regular (rotary) map M on a closed orientable surface. In fact, ΓB admits1 two such embeddings which are mirror images of each other such that their automorphism groups are isomorphic to G. In this case we may view Γ as obtained from M by truncation: cutting off each corner and then removing the edges in the triangles thus produced. In particular, let M be the tetrahedron and let G = A4 act on the vertices of M in its natural action. Then Construction 3.1 applied with Σ the underlying graph of M gives rise to Γ (Σ ) = 6 · K2 as shown in Fig. 1. 3.2. Flag graphs of 4-valent regular maps Next we construct four infinite families of graphs that arise in case (c) of Theorem 1.1 with m = 1. The constructions take as input a 4-valent regular map M with automorphism group G = Aut(M ) so (σ ) that the underlying graph Σ of M is G-symmetric and, for σ ∈ V (Σ ), Gσ ∼ = D8 . The output of = GΣ σ Construction 3.2 involves incident vertex–face pairs of M of the form (σ , h) where σ is a vertex and h is a face incident with σ . Construction 3.2. Let M be a regular map on a closed surface such that its underlying graph Σ has valency four, and let G = Aut(M ). For each edge {σ , σ 0 } of Σ , let f , f 0 denote the faces of M such that {σ , σ 0 } is on the boundary of both f and f 0 . Let oppσ (f ) and oppσ (f 0 ) be the other two faces of M incident with σ and opposite to f and f 0 respectively, and define oppσ 0 (f ) and oppσ 0 (f 0 ) similarly. Define four graphs Γ1 (M ), Γ2 (M ), Γ3 (M ), Γ4 (M ) with vertices the incident vertex–face pairs of M and adjacency defined as follows (where ∼ means adjacency): for each edge {σ , σ 0 } of Σ , (σ , f ) ∼ (σ 0 , f ) and (σ , f 0 ) ∼ (σ 0 , f 0 ) in Γ1 (M ); (σ , f ) ∼ (σ 0 , f 0 ) and (σ , f 0 ) ∼ (σ 0 , f ) in Γ2 (M ); (σ , oppσ (f )) ∼ (σ 0 , oppσ 0 (f )) and (σ , oppσ (f 0 )) ∼ (σ 0 , oppσ 0 (f 0 )) in Γ3 (M ); (σ , oppσ (f )) ∼ (σ 0 , oppσ 0 (f 0 )) and (σ , oppσ (f 0 )) ∼ (σ 0 , oppσ 0 (f )) in Γ4 (M ). Let B (M ) = {B(σ ) : σ ∈ V (Σ )}, where B(σ ) = {(σ , f ) : σ incident with f }. The following lemma shows that the graphs produced by Construction 3.2 have the required properties. Lemma 3.3. Let M, Σ , G be as in Construction 3.2 and let Γ = Γi (M ) be as defined there, where 1 ≤ i ≤ 4. Then Γ is a G-symmetric graph of valency two whose vertex set admits B (M ) as a G-invariant

1 Details may be obtained from the authors.

366

C.H. Li et al. / European Journal of Combinatorics 31 (2010) 362–367

partition such that k = v − 2 = 2, ΓB ∼ = Σ , and Γ B(σ ) = C4 is simple. Moreover, for adjacent blocks B(σ ), B(τ ) ∈ B (M ), Γ [B(σ ), B(τ )] = 2 · K2 . Proof. Since M is a regular map, G = Aut(M ) is transitive on the vertices of Γ and B (M ) is a Ginvariant partition of the vertex set of Γ . Since the underlying graph Σ of M is of valency four, the parts of B (M ) have size v = 4 and a typical part is of the form B(σ ) = {(σ , fi ) : 1 ≤ i ≤ 4}, where f1 , f2 , f3 , f4 are the faces of M surrounding σ . Let τi , 1 ≤ i ≤ 4 be the vertices of M adjacent to σ such that τi−1 and τi are incident with the face fi , where subscripts are taken modulo 4. If Γ = Γ1 (M ) then (σ , fi ) is adjacent to (τi−1 , fi ) and (τi , fi ) only, and hence Γ has valency two. Similarly if Γ = Γ2 (M ) then (σ , fi ) is adjacent to (τi−1 , fi−1 ) and (τi , fi+1 ) only, and again Γ has valency two. In either case Γ [B(σ ), B(τ1 )] consists of two edges, namely {(σ , f1 ), (τ1 , f1 )} and {(σ , f2 ), (τ1 , f2 )} for Γ1 (M ), and {(σ , f2 ), (τ1 , f1 )} and {(σ , f1 ), (τ1 , f2 )} for Γ2 (M ), and hence k = 2 and Γ [B(σ ), B(τ1 )] = 2 · K2 . Moreover, Γ B(σ ) is a cycle C4 , namely ((σ , f1 ), (σ , f2 ), (σ , f3 ), (σ , f4 ), (σ , f1 )) in both cases, and ΓB ∼ = Σ via the mapping B(σ ) 7→ σ . Since M is a regular map, there exists g ∈ Gσ which fixes f1 , interchanges τ1 and τ4 , and interchanges f2 and f4 . Thus g interchanges the two vertices adjacent to (σ , f1 ) in both cases, so Γ is G-symmetric. Similarly one can verify that all statements hold for Γ = Γ3 (M ) or Γ4 (M ). Each of Γ1 (M ), Γ2 (M ), Γ3 (M ) and Γ4 (M ) in Construction 3.2 is a union of cycles since it has valency two. For example, Γ1 (M ) ∼ = s · Ct and each face of M gives rise to a cycle of Γ1 (M ), where t is the face length and s the number of faces of M. For the octahedron M one can check that Γ1 (M ) ∼ = 8 · C3 , Γ2 ( M ) ∼ = 4 · C6 , Γ3 (M ) ∼ = 6 · C4 and Γ4 (M ) ∼ = 4 · C6 . 3.3. An explicit group theoretic construction Finally, we give a Sabidussi coset graph construction (see e.g. [6]) for an infinite family of graphs that satisfy part (b) of Theorem 1.1 with m = 1. Given a group G, a core-free subgroup H of G and a 2-element g such that g 6∈ NG (H ) and g 2 ∈ H ∩ H g , the coset graph Cos(G, H , HgH ) is defined to have vertex set [G : H ] = {Hx : x ∈ G} such that Hx, Hy are adjacent if and only if xy−1 ∈ HgH. It is known, see for example [6], that Cos(G, H , HgH ) is G-symmetric and is connected if and only if hH , g i = G. For a subgroup L < H, let B = [H : L] = {Lh | h ∈ H }. For x ∈ G, let Bx = {Lhx | h ∈ H }, and let B = {Bx | x ∈ G}. Then B is a G-invariant partition of [G : L]. Further, we have the following link between the two coset graphs. Lemma 3.4. Let Γ = Cos(G, L, LgL) and Σ = Cos(G, H , HgH ). Then Σ ∼ = ΓB . Proof. Define a one-to-one correspondence between [G : H ] and B by:

ϕ : Hx 7→ Bx ,

x ∈ G.

We claim that ϕ induces an isomorphism between Σ and ΓB . For any x, y ∈ G, we have (where ∼ means adjacency): Hx ∼ Hy

in Σ H⇒ yx−1 ∈ HgH

H⇒ H⇒ H⇒ H⇒ Bx ∼ By

yx−1 = h1 gh2

for some h1 , h2 ∈ H

1 −1 h− = g ∈ LgL 1 y(h2 x) 1 Lh2 x ∼ Lh− 1 y in Γ x y B ∼ B in ΓB .

in ΓB H⇒ Lh1 x ∼ Lh2 y

in Γ , for some h1 , h2 ∈ H

H⇒ h2 y(h1 x) ∈ LgL 1 H⇒ yx−1 ∈ h− 2 LgLh1 ⊂ HgH H⇒ Hx ∼ Hy in Σ . −1

Thus Σ ∼ = ΓB , as claimed.

Now we construct examples satisfying part (b) of Theorem 1.1.

C.H. Li et al. / European Journal of Combinatorics 31 (2010) 362–367

367

Construction 3.5. Let p be a prime such that p ≡ 1 (mod 16), and let G = PSL(2, p). Let H be a Sylow 2-subgroup of G. Then H = hai : hbi ∼ = D16 , ha4 , bi ∼ = Z22 , and NG (ha4 , bi) = S4 . There exists an 4 2 involution g ∈ NG (ha , bi) \ ha , bi such that g interchanges a4 and b. Let L = ha4 , bai ∼ = Z22 , and define

Σ = Cos(G, H , HgH ),

Γ = Cos(G, L, LgL).

Lemma 3.6. Using the notation defined above, the following all hold: (a) both Γ and Σ are G-symmetric, connected and of valency 4; (b) B is a G-invariant partition of V (Γ ) such that k = v − 2 = 2, Γ B = C4 and Σ ∼ = ΓB ; (c) for B = [H : L] and C = Bg ∈ ΓB (B), the induced subgraph Γ [B, C ] = K2,2 . Proof. It follows from the classification of the subgroups of G, see for example [7, pp. 417], that hH , g i is contained in no maximal subgroup of G. Thus hH , g i = G, and so Σ is connected. Moreover, since (a4 )g = b, it follows that b, a ∈ ha4 , ba, g i. Thus hL, g i = G, and so Γ is connected. By the definition, ha4 , b, g i ∼ = D8 , and H ∩ H g = ha4 , bi ∼ = Z22 . Hence Σ has valency 4. Since L is abelian, L ∩ Lg C L. Also L ∩ Lg is normalised by the involution g, and hence L ∩ Lg is normal in hL, g i = G. As G is simple, L ∩ Lg = 1, and so Γ is of valency 4. Part (a) now follows by Lemma 3.4. As above B is a G-invariant partition of V (Γ ) with parts of size v = |H : L| = 4. The stabiliser GB = H, and for C = Bg , we have GBC = GB ∩ GC = H ∩ H g = ha4 , bi. Label the vertex L of Γ as α . Then α ∈ B, Gα = L = ha4 , bai, and Gα ∩ GBC = ha4 i. The vertex β = α g = Lg lies in C ∩ Γ (α), and so 4

4

β a ∈ C ∩ Γ (α) and {β, β a } ⊆ C ∩ Γ (α). Also, since Gαβ = L ∩ Lg = 1, a4 does not fix β and hence 4 β 6= β a . Counting the numbers of edge of Σ and Γ , we conclude that there are exactly 4 edges of Γ between B and C . It follows that Γ [B, C ] = K2,2 and k = 2. This together with the fact that both Γ and Σ have valency 4 forces Γ B to be simple and isomorphic to C4 . Finally by Lemma 3.4, Σ ∼ = ΓB . This completes the proof of parts (b) and (c).

Corollary 1.3 follows from Lemma 3.6 immediately. Acknowledgements Research for this paper was supported by Discovery Project Grants DP0770915 and DP0558677 of the Australian Research Council. The second author was supported by Australian Research Council Federation Fellowship FF0776186. References [1] J.D. Dixon, B. Mortimer, Permutation Groups, Springer, New York, 1996. [2] A. Gardiner, R. Nedela, J. Širáň, M. Škoviera, Characterisation of graphs which underlie regular maps on closed surfaces, J. London Math. Soc. (2) 59 (1999) 100–108. [3] A. Gardiner, C.E. Praeger, A geometrical approach to imprimitive graphs, Proc. London Math. Soc. (3) 71 (1995) 524–546. [4] M.A. Iranmanesh, C.E. Praeger, S. Zhou, Finite symmetric graphs with two-arc transitive quotients, J. Combin. Theory Ser. B 94 (2005) 79–99. [5] C.H. Li, C.E. Praeger, S. Zhou, A class of finite symmetric graphs with 2-arc transitive quotients, Math. Proc. Cambridge Philos. Soc. 129 (2000) 19–34. [6] C.E. Praeger, Finite transitive permutation groups and finite vertex transitive graphs, in: G. Hahn, G. Sabidussi (Eds.), Graph Symmetry (Montreal, 1996), in: NATO Adv. Sci. Inst. Ser. C, Math. Phys. Sci., vol. 497, Kluwer Academic Publishing, Dordrecht, 1997, pp. 277–318. [7] M. Suzuki, Group Theory I, Springer, New York, Berlin, 1986. [8] S. Zhou, Constructing a class of symmetric graphs, European J. Combin. 23 (2002) 741–760. [9] S. Zhou, Almost covers of 2-arc transitive graphs, Combinatorica 24 (2004) 731–745; Combinatorica 27 (6) (2007) 745–746 (erratum).